Project Summary Medulloblastoma (MB) is the most common malignant brain tumor in children. Although aggressive treatments have improved outcomes, many MB patients still die of their disease, and survivors suffer severe long-term side effects from the therapy. Thus, an increased understanding of the disease and improved therapeutic approaches are critical. MB is comprised of four subgroups ? WNT, Sonic hedgehog (SHH), Group 3 and Group 4 ? that differ in terms of genetics, epigenetics, clinical characteristics and outcomes. Although we have some mechanistic understanding of WNT, SHH and Group 3 tumors, the genes and pathways that underlie Group 4, the most prevalent form of the disease, remain a mystery. Identifying oncogenic drivers and using them to create animal models of Group 4 MB is critical for development and testing of new therapies. Our analysis of the genome and epigenome of >400 human Group 4 MBs has identified single nucleotide variants (SNVs) as well as chromosomal copy number alterations (CNAs). While these studies have generated lists of candidate drivers, none of these has been tested or proven necessary or sufficient to initiate tumor growth. We hypothesize that perturbing expression of candidate genes or CNAs in neural progenitors will generate models of Group 4 MB, thereby validating their functional importance. To test this hypothesis, we propose to knock out or overexpress genes in murine and human (iPS-derived) neural progenitors and evaluate tumorigenic potential in vivo. These studies will create robust models of Group 4 MB that can be used to study tumor biology and to identify novel approaches to therapy.